Effects of computer administration upona tree drawing projective technique

Item Type text; Thesis-Reproduction (electronic)

Authors Pearce, Stewart, 1954-

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this materialis made possible by the University Libraries, University of Arizona.Further transmission, reproduction or presentation (such aspublic display or performance) of protected items is prohibitedexcept with permission of the author.

Download date 15/06/2018 23:52:01

Link to Item http://hdl.handle.net/10150/277283

INFORMATION TO USERS

The most advanced technology has been used to photograph and

reproduce this manuscript from the microfilm master. UMI films the

text directly from the original or copy submitted. Thus, some thesis and

dissertation copies are in typewriter face, while others may be from any

type of computer printer.

The quality of this reproduction is dependent upon the quality of the

copy submitted. Broken or indistinct print, colored or poor quality

illustrations and photographs, print bleedthrough, substandard margins,

and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete

manuscript and there are missing pages, these will be noted. Also, if

unauthorized copyright material had to be removed, a note will indicate

the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by

sectioning the original, beginning at the upper left-hand corner and

continuing from left to right in equal sections with small overlaps. Each

original is also photographed in one exposure and is included in

reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced

xerographically in this copy. Higher quality 6" x 9" black and white

photographic prints are available for any photographs or illustrations

appearing in this copy for an additional charge. Contact UMI directly

to order.

University Microfilms International A Bell & Howell Information Company

300 North Zeet) Road. Ann Arbor. Ml 48106-1346 USA 313 761-4700 800 521-0600

Order Number 1340261

Effects of computer administration upon a tree drawing projective technique

Pearce, Stewart, M.A.

The University of Arizona, 1990

Copiyright ©1990 by Pearce, Stewart. All rights reserved.

U M I 300 N. Zeeb Rd. Ann Arbor, MI 48106

BPPBCfS OP COMPUTES ADMIIII8TRATIOH UPON A

TREE DRAWIMG PBOJECTIVB TECHNIQUE

by

Stevact Pearce

Copyright (C) Stevart Pearce 1990

& Thesis Submitted to the Pacalty o£ the

DEPARTMENT OP ART

In Partial Pulfillaent of the Requirements Por the Degree of

MASTER OP ART8

In the Graduate College

THE UMIVRRSITY OP ARIZONA

19 9 0

2

STATEMENT BY AUTHOR

This thesis has been subaitted in partial fulfillnent of re­quirements for an advanced degree at The University of Arizona and is deposited in the University Library to be Bade available to borrowers under rules of the Library.

Brief quotations froa this thesis are allowable vithout special permission, provided that accurate acknowledgement of source is Bade. Requests for permission for extended quotation froa or reproduction of this aanuscript in whole or in part say be granted by the copyright holder.

SI

APPROVAL BY THESIS DIRECTOR

This thesis has been approved on the date shown below:

J Ass i stair

Carrigan Professor of Art

/ffo Date

3

ACKNOWLEDGEMENTS

Sincere gratitude is extended to Professors Lynn Galbraith,

Darrell Sabers, Jean Rush, and f. Dvaine Greer who instructed and

advised ae. Special thanks is given to ay advisor and thesis director,

Sister Jeanne Carrigan for her inspiration and patience. I aa indebted

to Professor Jon Sharer, who introduced ae to the vonders of coaputer

graphics. Recognition is given to the Right Answers Group, producers

of the coaputer prograa that allowed ae to generate the display

sequences used in Coapute-A-Tree and Coaaodore-Aaiga, producers of the

aachine that Bade this study possible.

I aa especially grateful to Carol J. Corsica, who gave advice,

proof read, and provided invaluable support. Coapute-A-Tree and this

thesis are dedicated to the aeaory of Joyce Kilaer for whoa trees grow

and poppies blossoa.

4

TABLE OP COMTBNT3

Page

ABSTRACT 6

CHAPTER

I INTRODUCTION 7

Rationale 8

Mull Hypothesis 10

Thesis Questions 10

Bxperiaental Design 11

Subjects 11

Definitions 12

Liaitations 15

Suauaary 17

II LITERATURE REVIEW 18

House-Tree-Person Test 18

H-T-P Reliability 22

H-T-P Validity 23

Der Bauatest 26

Tree Draving Test 27

Kinetic-House-Tree-Person Technique 28

Diagnostic Draving Series 29

Coaputer Projective and Therapy Techniques 30

Suaaary 31

III PROCEDURES 34

Design Guidelines 34

Equipaent Used 36

5

TABLE OP CONTENTS Continued

Page

Conpute-A-Tree Iaages 36

The Pencil and Paper Instruaent 43

The Scoring Systea 43

C&T Reliability 46

C&T Validity 47

Suaaary 52

IV RESULTS 53

Correlations 54

Variance Vithin Groups 55

Variance Across Groups 55

The Effect of Time 56

Specific PDQ Stateoents 57

Suutary 60

V DISCUSSIOH 62

Conclusions 65

APPENDIX A: CAT QUANTITATIVE SCORING SYSTEM 67

APPENDIX B: POST DRAWING QUESTIONNAIRE (PDQ) 70

APPENDIX C: FACSIMILES OP SELECTED CAT MENUS 73

APPENDIX D: HOUSB-TRBB-PBRSON POST DRAVING INTERROGATION (PDI). . 75

APPENDIX E: DATA ANALYSIS 77

REFERENCES 86

ABSTRACT

Projective tree drawing techniques are used by clinicians and

therapists to assess the personality and eaotional state of patients.

This study compares the results of adainisterlng a computerized

projective tree drawing technique with the results obtained by a pencil

and paper counterpart. Both techniques are based upon the tree drawing

component of the House-Tree-Person technique and related tests. Vith

Coapute-A-Tree, subjects created tree pictures fro* a aenu of

preselected iaagery while subjects taking the conventional fora of the

technique produced spontaneous tree drawings. A post-drawing

questionnaire (PDQ) eaploylng a Likert scale was used to aeasure

subjects' attitudes regarding their tree images. The Bean score for

coaputer rendered trees was higher than the aean score for

conventionally rendered iaages. Siailarities were found between

responses to the Iaages obtained through the two foras of

adainistration.

CHAPTER I: INTRODUCTION

Figure drawings have been used by clinicians as diagnostic tools

for assessing the personality and the emotional state of patients since

art production was first recognized in the 1850s as a means of

expressing both conscious and unconscious personality tendencies

(Hammer, 1980). Projective draving techniques were developed in the

late nineteenth century as a systematic means of observing

characteristics of affect (Burns, 1987). Buck (1948), Burns (1987),

and other authors of projective tests believed that the images created

by a patient contained observable indications of that patient's

physical and emotional needs as veil as indications of psychological

defense against the anxiety causing nature of that patient's feelings,

motivations, and behaviors (Rabin, 1968).

In 1948, Buck developed the House-Tree-Person (H-T-P) test as an

instrument for measuring intelligence and personality through

figurative draving. Since that time, test designers have questioned

the theoretic rationale, the reliability, and the validity of

projective draving techniques including H-T-P (Anastasi, 1988).

Killian (1984) criticized Buck's H-T-P technique by stating that

it allovs unacceptable subjectivity on the part of examiner

interpretations of projective dravings. Buck (1985) stated, "There is

almost no statistical proof of the validity of the quantitative scoring

points and their interpretations" (p. 164). Nevertheless, Buck

believed that H-T-P vas valid based upon the empirical evidence present

in the observations of clinicians and insightful test subjects of H-T-

P. Although Killian (1984) criticized H-T-P, he reported that

8

clinicians viewed projective techniques with enthusiasm.

Instruments that use draving as tools for measuring intelligence

and personality are still being developed. Naglieri (1988) developed

Draw A Person; A Quantitative Scoring System (DAP) as a modernized,

recently normed, objective human figure draving test that obtains

estimates of intelligence in a non-threatening manner through a process

less influenced by language variables. In his introduction to DAP,

Naglieri suggested that DAP's quantitative scoring system does not

alter projective interpretations of DAP's human figure drawings using

other evaluative systems (1988). Although Naglieri designed DAP as a

test of intelligence rather than of psychological projection, he did

not hesitate to point out that DAP drawings could be interpreted using

other projective systems.

Cohen, Hammer, and Singer (1988) felt that the plethora of

competitive projective drawing techniques led to a "horrible spravl of

mostly incompatible information". Cohen (et al., 1988) designed the

Diagnostic Drawing Series (DDS) as an art therapy tool that

systematically relates the graphic images of patients to Diagnostic and

Statistical Manual of Mental Disorders (DSM-III-R, 1987) diagnoses.

Levy and Barowsky (1986) studied the effect of using a computer to

administer the Goodenough-Harris Draw-A-Man Test. Their results

suggest that computer technology might be used to administer projective

drawing techniques that, in the past, relied on pencil and paper

administration (Levy and Barowsky, 1986).

Rationale

Conditions for test administration including standardization and

accuracy have been found to improve vhen computer administration is

employed (Brovn, 1984; Committee on Professional Standards and

Committee on Psychological Tests and Assessment, 1986). Computer

administration of projective techniques might provide a more objective

and accurate means of data collection for relating patients' graphic

images to DSM-III-R (1987) diagnoses. In this respect, computerizing a

projective tree draving technique might eliminate subjectivity vhen

scoring and interpreting patients' vorks. Computer administration

might be used to satisfactorily establish validity and reliability in

H-T-P and related techniques. Such techniques could be adapted fox

special and handicapped populations.

I created and programed Compute-A-Tree (CAT) to test computer

administration as an alternative to pencil administration of projective

tree draving techniques. Tree images created by subjects using the CAT

computer program vere quantitatively compared vith spontaneous dravings

created by subjects using pencils and paper. I also developed the

quantitative scoring system that vas used to assess tree images. The

quantitative scoring system appears in Appendix A. Using the scoring

system, I assigned numeric values to specific characteristics that

appeared in both computer and spontaneously dravn tree images. I

developed a post draving questionnaire (PDQ) In order to ascertain

subjects' attitudes tovards pencil and paper dravings and computerized

tree images. The PDQ appears in Appendix B. The PDQ measured

subjects' attitudes about their dravings and computer images using a

summative or Likert scale. The five responses In the PDQ Likert scale

vere assigned numeric values to facilitate the comparisons.

10

Null Hypotheses

The following statements represent the null hypotheses of this

study:

HOI There will be no quantitative difference between CAT tree

iaages and drawn tree iaages.

H02 There will be no quantitative difference between subjects'

attitudes regarding the coaputer generated tree iaages and subjects'

attitudes regarding the spontaneously drawn tree iaages.

H03 Posttest scores for either group will not be influenced by

the aaount of tiae between two administrations of the projective tree

drawing technique.

Thesis Questions

Three thesis questions will be considered in the course of testing

the null hypotheses:

1. The first question considers subjects attitudes about the

coaputer iaages they created with preselected iaages. Did subjects

express favorable attitudes about the trees they created on the

coaputer using preselected iaagery? Did subjects have a different

attitude for hand drawn trees?

2. The second question relates to the concept of projection. CAT

was developed as a coaputer adainistered projective technique. When

the subjects of Buck's (1948) study drew trees, they unconsciously

related the tree iaages to theaselves in such a way that the tree

iaages actually becaae syabols of the subjects. Did subjects responses

on the CAT PDQ tend to indicate that they relate to the coaputer iaages

in the saae way that they relate to spontaneously drawn iaages?

11

3. The third question deals vith the way imagery vas presented in

the computer simulation. Subjects constructed trees on the computer by

choosing £rom alternative preselected images vhen they vere prompted to

do so. The menu of alternative images and the prompts formed a

decision-inducing set. Did this computer presented set influence

subjects1 responses?

Experimental Design

A modified pretest-posttest control group design vas used to test

the null hypotheses vith a significance level of alpha = .05. The

modified pretest-posttest model assumed this form:

R X 01 02

R 03 04

where X is the computer administration itself. In classic pretest-

posttest design, the experimental group receives treatment after both

groups are pretested (Campbell and Stanley, 1963). Since I compared

administrations of the projective technique, the computer

administration serves as the experimental treatment. The scores of the

computer administration became the first observations of the

experimental group.

The pencil and paper version of the projective technique and the

PDQ served as the pretest for the control group and the posttest for

both the experimental and control groups. Whenever possible, control

and experimental subjects vere administered the projective tree draving

technique in pairs. I gave the post-test to mixed groups of

experimental and control subjects.

Subjects

12

In classic pretest-posttest design, randomization of subjects

insures that no initial bias occurs betveen groups (Campbell and

Stanley, 1963). Sixty-eight volunteer subjects vere randomly assigned

to the experimental and control groups by the flip of a coin. Seven

subjects vere male and vere deleted from the sample used for the

quantitative analysis in order to keep the sample homogeneous. Three

subjects from the experimental group vere deleted from the study

because they had college level art training, a variable that vould

compromise results (Bieliauskas and Bristov, 1959). There vere 27

experimental and 31 control female subjects for this study.

All subjects vere elementary teacher candidates at a southvestern

university vho indicated that they had less than a year of formal art

training. All subjects vere currently enrolled in an art methods class

for elementary teacher candidates. Subjects ranged in age from about

22 to 40 years. Subjects also came from diverse ethnic backgrounds.

Experimental group subjects indicated that they had varying degrees of

computer experience. Four subjects indicated that they used the

computer extensively as a vord processor, nine indicated that they

never used a computer and one described herself as having "computer

phobia". The remaining subjects indicated that they used the computer

or a vord processor occasionally, usually vhen college vork required

it.

Definitions

The folloving discussion presents key terms and concepts found in

subsequent chapters of this study:

Projective drawing techniques. Projective techniques are used by

13

clinicians and act therapists as aeans of stimulating and facilitating

the expression of imagination in order to reveal disguised or

unconscious inclinations of normal persons (Rabin, 1968). The purpose

of projective drawing techniques is disguised, to reduce the chance

that responses vill be deliberate to create an impression rather than

be true reactions of affect.

Drawings created through projective techniques serve as

expressions of affect in that they are projections of conscious and

unconscious feelings, motivations, and behaviors as these relate to

personality (Hammer, 1968). As part of the theory behind H-T-P, Buck

(1948) postulates that subjects communicate emotion through the details

included in drawings as well as through the sequence and time elements

used to draw those details. Projective techniques consider the

summation of personality traits in a global appraisal rather than

measuring specific traits (Anastasi, 1988).

Computer administered projective technique. CAT is a computer

administered projective technique because the computer presents the

visual test stimulus and records responses in the computer data base.

Computer administration differs from computer assistance in that

the computer is the sole instrument used for instruction or testing

purposes (Kearsley, 1986). As an example, Miller's (1986) Online

Computer-Assisted Rorschach Inkblot Test had subjects handle the

traditional inkblot cards while recording their responses using the

computer. A computer administered version of the same test would have

inkblot images presented through the computer monitor.

CAT menu. The CAT program presented and recorded subjects'

14

responses to menus of images. A facsimile of selected CAT menus

appears in Appendix C. During the experiment, one menu at a time

appeared on the computer monitor, offering subjects alternative pre­

programed images. The selection of preselected images served as the

stimulus for projection in CAT.

CAT prompt. CAT prompts were verbal statements that asked

subjects to choose between the alternative images by typing a single

letter of the alphabet. Prompts served as the stimulus for selecting

images, therefore the prompts were also a stimulus for projection in

CAT.

Decision-inducing set (PIS). CAT menus and prompts appeared

together so that subjects were presented with a situation that requires

them to choose from alternative images. Subjects could change their

choice by electing to return to a given situation or DIS.

Pixel. The square picture elements of a computer monitor are

referred to as pixels. CAT was displayed using a low resolution pixel

array of 320 wide by 200 high in order to conserve chip RAM. Pixels

are noticeable at low resolution when curvilinear lines are rendered.

Pixels are less noticeable in curvilinear lines that are displayed at

high resolution with an array of 640 wide by 400 high.

Screen buffer. Computer screen buffers served as storage areas in

the computer memory for computer images. CAT images were created using

a commercially available "paint" computer graphics program. Images

were stored in interchange file format (IFF) on a 3.5 inch micro-floppy

disk. To sequence displays of the CAT images, IFF images were

loaded into one of fifty screen buffers as a result of subjects' key

15

board input.

Only one screen buffer could be loaded into the computer memory at

a time but they were sequenced, one after the other, vith double

buffering. One buffer could overlap a second buffer in a stencil

fashion. This overlapping is hov tree top and tree bottom images vere

combined to make a complete tree on the computer.

Chip RAM. RAM is the abbreviation for "random-access memory".

Chip RAM is the amount of display memory that display buffers can

reside in. The Commodore Amiga computer disk drive used for

administering CAT had 400,000 bytes of display memory. A two color

image displayed on the Amiga computer monitor at lov resolution

requires 8,000 bytes of memory.

Limitations

The CAT program could produce approximately 2,280 tree images

using a combination of 40 computer generated tree trunks and 57

computer tree tops. Increasing the number of tree trunk and tree top

images vas not possible due to limitations in the computer memory.

The PDQ did not measure affect or emotion in the same way true

projective techniques do. Buck's (1948) H-T-P technique includes both

a drawing phase and a "post-draving interrogation" (PDI) where the

examiner asks open ended questions about the drawings (p. 328). The

CAT PDQ was designed for quantitative analysis. The CAT PDQ statements

were based upon Buck's PDI, however the PDQ statements were specific

rather than open ended. Although the CAT design was based upon non­

verbal projective tree drawing techniques, computer administration and

the use of the written PDQ made CAT a verbal test.

16

Mo atteapt vas Bade to natch subjects' ages for experimental and

control groups. No atteapt vas Bade to Batch subjects' ethnicity for

the experiaental and control groups. Subjects' prior coaputer

experience vas not considered.

Subjects' dravings, CAT results, and PDQ responses vere neither

psychologically nor symbolically interpreted by the author of this

study. Buck (1948) admitted that H-T-P dravings can be misinterpreted

when a patient relies on previous siailar draving experiences or

training and creates a stereotypical image. According to Buck (1948,

1985), vhen patient dravn stereotypical iaages are discovered, they

should only be scored qualitatively. The projective tree draving

technique presented in this study provided no aethod for ascertaining

if subjects drev stereotypical images.

The modified pretest-posttest design of the experiment created a

situation vhere test-retest reliability could be considered. Scorer

reliability vas not considered an important factor for CAT because the

preselected computer images vere assigned numeric values prior to test

administration. The computer documented subject responses and the

scores as they occurred. Spontaneous tree dravings vere scored using

the same criteria established for coaputer images. The computer images

served as samples for assigning draving scores in the same vay that the

images included in Buck's (1948) handbook served as samples for scoring

the results of H-T-P.

The scores obtained froa CAT and the pencil and paper counterpart

vere not correlated vith scores obtained froa other projective

techniques. In this experiment, the internal consistency of the PDQ

17

vas coaputed by correlating each response vith the total score.

Internal consistency vas the only fora of construct related validity

considered for this experiment.

Summary

Projective draving techniques are used by clinicians and

therapists to assess the personality and emotional state of patients.

Test instruaents that use draving to neasure personality have been

criticized because of questionable validity and reliability. Computer

technology can be used to administer and score clinical instruments

resulting in aore accurate and better standardized administration.

CAT, an original computer administered tree draving projective

technique, is described in this study. Tested vere the null hypotheses

that there vas no quantitative difference in tree image scores betveen

the tvo groups, that there vas no quantitative difference in

attitudinal responses to tree images as obtained using a Likert scale,

and that time betveen treatments had no effect on post-test scores for

the tvo groups. A modified pretest-posttest control group design vas

used for this study.

The results of testing CAT on a sample population of 27 female

experimental subjects are presented. The experimental results are

compared vith results obtained from 31 female control group subjects

vho vere administered a pencil and paper version of CAT.

The folloving chapter provides background information regarding

five projective tree draving techniques. The literature revieved

served as the basis for developing CAT.

CHAPTER II: LITERATURE REVIEW

According to Buck (1948, 1985) tree drawing is an excellent

stimulus for eliciting projection of affect as the tree appears to

represent the subject's unconscious picture of himself or herself in

relation to the subject's general psychological field. The tree iaage

is one of the first images drawn by children and one of a few images

that adults feel competent to draw (Cohen, et al., 1988). Specific

elements within the tree drawing may indicate the subject's

subconscious view of self development, psychosexual maturity, contact

with reality and intra-personal balance (Buck, 1948).

House-Tree-Person Test

Buck developed H-T-P in 1948 from tests he originally conceived as

part of an intelligence Inventory. In doing so, Buck followed a

tradition established by Goodenough's Draw-A-Man (D-A-M) test,

published in 1926 as an intelligence measure. Goodenough's D-A-M test

derived its scores solely from the characteristics that appeared in

drawings (Harris, 1963). Buck (1948) derived his scores from a

combination of inventoried drawing characteristics and verbal responses

to the clinician's questions.

In the achromatic phase of H-T-P, the examiner gives the subject

labeled, white, 7x8 1/2 inch paper and a pencil. The examiner

states, "I want you to draw me as good a picture of a house (or tree or

person) as you can" (Buck, 1948, 1985). Using a stopwatch, the

examiner times how long the subject takes to produce the drawings of

the house, tree, and person. The examiner also records the order that

the subject used when drawing specific parts of the house, tree, and

19

person as veil as any spontaneous consents aade by the subject.

After the subject coapletes the three requested drawings, the

exaainer asks pre-specified questions about the drawings using the

format of the PDI. The PDI provides the subject with the opportunity

to verbally project into the graphic descriptions. The PDI also

provides the exaainer with the opportunity to clarify the aeaning of

aabiguous graphic descriptions. PDI questions that related to tree

iaages are included in Appendix D.

At the coapletion of the PDI, the exaainer asks the subject to

produce a second chroaatic drawing of a house, tree, and person using

wax crayons in place of the pencil. The exaainer scores the dravings

and verbal responses of the subject using the scoring tables and the

saaple dravings provided in the H-T-P test aanual.

The drawings are scored based upon the specific characteristics

that appear in each, however specific characteristics Bay have aore

than one interpretation (Buck, 1985). Misinterpretation of patient's

drawings occurs when subjects draw stereotypical iaages or when

subjects becoae defensive about their drawing capability and are at a

loss when asked to produce a drawing of a house, tree, or person even

though they are assured that H-T-P is not a test of artistic skill

(Buck, 1948). The exaainer views dravings in light of what subjects

have said during the interview.

The exaainer considers the content of PDI verbal responses and the

conventionality of those responses froa average to unusual,

unconventional, or pathological (Buck, 1985). The exaainer also

considers the subjectivity and aultiplicity of verbal references of

subject responses. Organization and consistency o£ drawings and the

degree to which the subject illustrates things that are unpleasant to

the subject are noted by the exaainer.

Using the scoring points, the exaainer derives an 1Q figure for

the percentage of raw G, the net weighted score, the weighted "good"

score, and the weighted "raw" score. The exaainer assesses the

subject's personality by considering the drawings, the subjects verbal

responses to PDI questions and the subject's spontaneous responses

(Buck, 1985).

Buck (1948) developed noras for the quantitative scoring systea by

adainistering H-T-P to institutionalized patients who were aentally

deficient with functional or organic pathology. Buck conducted

additional studies with school children, college students, and non-

institutionalized adults.

In these initial studies, Buck postulated that subjects' drawings

contained iaagery that were responses of affect. Buck's observation

seeaed to confira reports by Goodenough and other clinicians who stated

that personality factors surfaced in drawn works when children were

administered the D-A-M test (Hammer, 1968, 1980).

Buck originally stressed that H-T-P represents "a valid aeasure of

adult intelligence, despite its restricted and unconventional approach

to such aeasureaent" (1948, p. 321). By including details, indications

of spatial relationships, and concept foraations in their H-T-P

drawings, subjects could reveal the presence of otherwise undetected

intellectual ability (Buck, 1948).

In his revision of H-T-P, Buck (1985) clarified the intent of his

instrument. He stated that H-T-P represents a nethod of assessing

intellect "in a situation deliberately designed to activate non-

intellectual aspects of the personality which enhance or diminish

efficiency of intellectual function" (Buck, 1985, p. 3). Buck (1985)

suggested that one can aeasure affect by considering its influence upon

intelligence.

Although Buck (1985) adaitted that his scoring systea and drawing

interpretations were not statistically proved, he did conduct

standardization studies for H-T-P. Buck (1948) selected 140 subjects

to serve as the saaple size for developing noras for the quantitative

scoring systea. Buck (1948) divided the saaple subjects into seven

groups of 20 persons each. According to Buck, each division

represented an intelligence level (iabecile, aoron, borderline, dull

average, average, above average, and superior). Subject assignaent to

intelligence levels vas based upon each patient's "clinically

deaonstrated level of intellectual function. . .and not a score on one

or aore standard intelligence tests" (Buck, 1985, p. 8). Buck (1948)

obtained his results by using single exaainatlons for subjects of

average and belov average intelligence and he used group exaainations

for subjects of above average intelligence.

Buck conducted qualitative standardization studies using 150

subjects (1948). Through this second set of studies, Buck identified

variations in dravings of noraal and aaladjusted subjects. Buck (1948)

assigned the subjects of this second study to one of eight groups of

personality aaladjustnent. Buck included no discussion regarding the

criteria used for assigning this second set of subjects to their

22

respective groups.

H-T-P Reliability

The presence and influence of the exaainer on subjects of H-T-P,

the effects of art training on subject's H-T-P drawings, and the size

of H-T-P drawing fores are all factors that cause variations in the

outcome of H-T-P adainistration. Each of these factors can

demonstrably change H-T-P scores and the statistical validity and

reliability of Buck's projective technique. These factors can also

effect of H-T-P interpretations. The following discussion provides

information that is helpful in gaining insight into H-T-P as a

projective technique. The discussion is also germane to the design and

administration of the CAT instrument used in this study.

Cassel, Johnson, and Burns (1958) adainistered H-T-P to a

heterogeneous population of 130 adults in order to test the effect of

examiner presence on subject drawing responses. Their study found that

the presence of the examiner resulted in fewer interpretive features in

all three drawings with the tree drawing least affected. Drawing sizes

tended to increase when the examiner was not present during H-T-P

administration with the tree drawing being aost affected by exaainer

presence.

Bieliauskas and Bristow (1959) studied the effects of formal art

training upon H-T-P quantitative scores. Bieliauskas and Bristow

adainistered both the drawing component of H-T-P and an abbreviated PDQ

to 60 college students, half of whom were art majors. The results

indicated that art students score significantly higher on H-T-P drawing

and IQ aeasureaents than do non-art students.

23

Bieliauskas and Farraghec (1983) tested hov changing the H-T-P

drawing sheet size effected the resulting IQ score obtained by Buck's

test. They found that subjects handled formal drawing properties of

proportion and perspective vith greater success when smaller forms (4

1/2 x 5 1/2 inches) vere used in place of the standard drawing sheet (7

x 8 1/2 inches) advocated by Buck.

H-T-P Validity

Anastasi (1988) suggested that the elaborate quantitative scoring

systems used by some projective techniques are deceptive, creating the

illusion that the techniques are objective measures. DiLeo (1983)

observed that the analysis of drawings used as projective instruments

have failed to consistently provide correlations between drawings and

levels of mental and emotional maturity as measured by IQ tests and

personality instruments. Lack of correlation exists because

intangibles like expressions of feelings resist quantification.

Although internal consistency was the only form of construct related

validity considered in the CAT experiment, the following discussion of

is helpful for gaining insight into the validity of H-T-P as a

projective technique.

According to Killian (1984), the standardization research of

H-T-P is not sufficient to meet the criteria for acceptable methods of

test development and construction due to the small sample size. In

Killian's opinion, "Unless there is research support, the H-T-P may be

relegated to the position of only a therapeutic tool for art therapy or

be a limited diagnostic tool for non-verbal patients" (1984, p. 345).

Despite the small sample size of the original studies, Buck (1948)

24

did show that H-T-P appraises general intellectual function as veil as

non-intellectual factors. According to Buck (1985), the correlation

coefficient betveen H-T-P percentage of rav G IQ and the IQs obtained

by other tests vere as follovs: Otis Higher Examination, .41 (30

subjects tested); Stanford-Binet Forms L-M, .45 (26 subjects tested);

Wechsler-Bellevue Verbal IQ, .699, Performance IQ, .724, and Full IQ,

.746 (each vith 100 subjects tested). Although sample sizes for

comparing H-T-P to the Otis Higher Examination and the Stanford-Binet

Forms L-M vere small, the correlations are significant at p = .05.

Killian (1984) pointed out that studies conducted in 1970 by

Hellkamp and Johnson found nonsignificant correlations betveen H-T-P

IQs and those obtained vith the Wechsler Adult Intelligence Scale.

Using the H-T-P achromatic test and the Kuhlman-Anderson Intelligence

Test, Bieliauskas and Moens (1961) found that the H-T-P scoring system

could legitimately be used for group predictions of IQs but vas not

applicable for making individual predictions vhen considering the

tested population of 23 second graders and 40 fifth graders.

Wildman and Wildman (1975) conducted a survey of the diagnostic

validity of H-T-P vhen used singularly and in combination vith the

Bender Visual Gestalt Test (Bender-Gestalt). Six clinicians vere asked

to differentiate betveen the H-T-P and Bender-Gestalt tests of 10

patients and 10 nurses. Wildman and Wildman (1975) found that the

clinicians could not differentiate betveen the tvo populations at above

chance level (or 53% of the time) vhen using H-T-P alone or in

combination vith the Bender-Gestalt. Conversely, Wildman and Wildman

found that clinicians could differentiate betveen the tvo populations

vhen only Bender-Gestalt scores vere considered. Wildaan and Vildaan

concluded that the coebination of the tests did not increase the

accuracy of diagnostic predictions (1975). Wildaan and Wildaan drev

this conclusion from an exceedingly snail group of subjects. Killian

(1984) used Wildaan and Wildaan's survey to infer that incorporating H-

T-P in a test battery Bight "increase error and reduce the probability

of Baking an accurate diagnosis" (p. 344).

Research by Covden, Deabler, and Feamster (1955) indicated that

clinicians Bight be able to sake general predictions regarding

patients' eaotional adjustment using a battery of tests that include H-

T-P. Four clinicians vere able to differentiate between the 58

subjects of the control group, the group that required continued

hospitalization, and the group that vas judged as ready for hospital

discharge (Covden, et al., 1955). The study (Covden, et al., 1955)

found that clinicians could not predict levels of adjustaent but

changes in a patient's adjustaent could be predicted.

In 1972, Marzolf and Kirchner conducted a study to see if there

vere significant relationships betveen draving characteristics and

personality traits. Scores vere obtained froa 760 college students vho

vere given both H-T-P and the Sixteen Personality Factor Questionnaire

(16PF). Marzolf and Kirchner (1972) developed a list of 108

personality characteristics found in H-T-P dravings using Buck's

original test. This list served as the basis for coaparing H-T-P and

16PF test results. All of the 16PF traits vere associated vith soae

draving characteristics vith "tree branches droop" entering into five

relationships, "aore than. . .any other draving feature" (Marzolf and

26

Kirchner, 1972, p. 161).

In general, the data gathered by Marzolf and Kirchner (1972)

suggests that there are valid relationships between H-T-P draving

characteristics and personality. The study found that the neaning of

various draving characteristics varied with the sex of the subjects.

According to Marzolf and Kirchner (1972), such characteristics say be

associated with personality in a non-linear fashion. Findings by

Harzolf and Kirchner (1972) are significant to those who use projective

techniques because the findings suggest that the draving coaponent of

Buck's instrument say not reveal as such about patients as the PDI.

Case studies by Hammer (1980) and Nauaberg (1980) indicated that

projective draving techniques such as H-T-P are both useful and

valuable in the treataent of patients. Their findings reinforced

Buck's observation that the validity of projective draving techniques

nust be considered not only fron the statistical standpoint, but also

froa the empirical evidence presented by clinicians and therapists vho

use projective draving techniques.

Der Bauatest

Buck vas not alone in his observation that eleaents of a tree

draving can represent aanifestations of the subconscious. In 1949,

Koch published Der bauatest. a projective technique that asked subjects

to drav a fruit tree (Bolander, 1977). Koch, a Swiss vocational

consultant, aodified a tree test designed in 1928 by his senior

colleague, Enil Jucker. Koch conducted noraative studies on children

and then extrapolated his findings to other populations (Bolander,

1977).

27

Koch's projective technique provided the examiner vith a limited

inventory of drawing characteristics and an extensive list

of possible Meanings for each (Bolander, 1977). Koch referred to the

study of graphology when designing Per bauatest and the 1952 English

translation utilizes the specialized vocabulary of handwriting analysis

(Bolander, 1977).

Tree Drawing Test

Bolander (1977) provided a third projective tree drawing technique

based upon the unpublished works of Hungarian priest, Karoly Abel.

Like Per bauatest. Bolander's technique used the tree as the sole theae

for projective drawing. Unlike H-T-P, Bolander's instruaent eaployed

no required aaterial list. Color drawings were not elicited, although

subjects were allowed to use color if they wished. Bolander's

instruaent had no required adainistrative procedure. There were no

time constraints and no emphasis was placed on exaainer observations or

presence during test adainistration. Bolander used no structured PDI

and interviews were conducted only after the drawings were subjected to

analysis.

Bolander (1977) presented her tree drawing projective technique as

an alternative to H-T-P. She did not consider her technique to be a

proper measure of intelligence. Although Bolander tested her

instruaent on a population of 3,174 persons, she suggested that the

saaple population did not constitute a representative cross section of

the general population. College students constituted 43% of Bolander's

selected population while 34% were professional adults and 1% were

children and adolescents all of whoa were of "noraal and above noraal

28

intelligence. . .who aanaged their lives relatively veil" (Bolander,

1977, p.7).

Bolander (1977) provided syabolic aeanings for tree drawing

characteristics but she stressed that no single interpretation was

correct for any of the syabols identified in her projective technique.

Bolander claiaed that H-T-P was inappropriate as a projective technique

for the individuals represented by her selected population. To justify

her clala, Bolander cited her own case studies where extreaely gifted

individuals produced drawings that Bight indicate iabecility on the

part of the artist when scored using Buck's scoring system. Bolander

nade no claims regarding the validity and reliability of her tree

drawing test.

Kinetic-House-Tree-Person Technique

Burns (1987) used the structure of H-T-P to create the Kinetic-

House-Tree-Person technique (K-H-T-P). Burns suggested that coabining

the house, tree, and person eleaents in one drawing would result in a

drawing that reveals the dynaaics between the subject, the subject's

hoae life, and the subject's view of her or his physiologic and

psychologic developaent (1987).

Burns (1987) indicated that drawings should be interpreted by

considering Abrahaa Maslow's developaental aodel of aental growth.

According to Burns (1987), Haslow viewed aental developaent as a

process aotivated by a hierarchy of needs leading to self

actualization. K-H-T-P was created as a bridge between projective

techniques, developaental psychology, and the systeas of inner personal

exploration suggested by Maslow and influenced by Eastern religions and

29

philosophies (Burns, 1987).

K-H-T-P differs from H-T-P in that there is no proscribed PDI at

the conclusion of K-H-T-P. K-H-T-P interview questions inquire into

the visual and emotional nature of the images portrayed. Burns (1967)

described no actual scoring of subject's iconographic and verbal

responses. Burns provided an inventory of characteristics that may be

found in K-H-T-P drawings. Burns derived symbol interpretations both

from his own observations and those of Buck (1948), Hammer (1980), and

others.

Like Bolander, Burns (1987) provided graphic examples of house,

tree, and person types, however the validity of the symbolic

interpretations and the K-H-T-P test as a whole are described only in

terms of 48 specific case studies. In his criticism of K-H-T-P, Hammer

(1988) suggested that the data supporting Burns' interpretation

appeared to be selectively assembled. Burns (1987) consistently

reported that the K-H-T-P outperformed Buck's test in all 48 cases. In

combining the house, tree, and person images, more details are lost in

the images themselves as the patient attempts to relate the three

concepts to each other (Hammer, 1988).

Diagnostic Drawing Series

Cohen, Hammer, and Singer (1988) included a tree drawing component

in their Diagnostic Drawing Series (DDS). According to Cohen (et al.,

1988), DDS represented the first projective technique that was linked

through research with the Diagnostic and Statistical Manual of Mental

Disorders (DSM- III-R, 1987). DDS consisted of a "free" unstructured

drawing, a drawing of a tree, and a drawing using lines, shapes, and

30

colors in a vay that expressed the patient's emotions (Cohen, et al.,

1988).

Cohen (et al., 1988) performed a preliminary study of DDS using

239 subjects diagnosed in the categories of schizophrenia, depression,

and dysthymia. Each of the DDS dravlngs vas scored based on the

presence, absence, or quality of 36 characteristics (Cohen, et al.,

1988). Although the authors of DDS ventured to suggest that some

relationship exists between pictorial structures in patients' drawings

and patients' psychiatric diagnoses, the sample size limited the

conclusions that could be drawn from the study (Cohen, et al., 1988).

Computerized Projective and Therapy Techniques

Canoune and Leyhe (1985) discovered that subjects tended to

volunteer more information through computer interview than through

human interview. In a pilot study using 36 subjects, Selmi, Klein,

Greist, Sorrell, and Erdman (1990) found that patients receiving

computer administered cognitive-behavior therapy for depression

improved as much as patients receiving therapy from a clinician while

both therapy groups improved significantly from the control group that

received no therapy.

Although the results of each of the studies may be due, in part,

to the novelty of the computer application, each study suggests that

computer technology can be successfully used as a diagnostic and

therapeutic tool in clinical settings. In addition, the computer

treatment equaled conventional treatment.

Levy and Barowsky (1986) compared computer administration of the

D-A-M with standard administration. Their study was probably the first

instance where a projective drawing technique vas computerized. A

heterogeneous population of 40 high school students vas administered D-

&-M. Levy and Barovsky (1986) found that the 20 subjects vho took the

standard form of D-A-M included more body and clothing parts than did

the 20 subjects vho took the computerized form of D-A-M. Girls scored

higher on the traditional D-A-M while boys displayed no significant

difference betveen traditionally obtained and computer obtained scores

(Levy & Barovsky, 1986).

Killian (1984) stated that computerizing H-T-P is impossible

because of the detailed, qualified, and ambiguous scoring criteria and

the continued need for comparing subject responses vith H-T-P manual

descriptions. Some of these limitations may be overcome by providing

subjects vith predefined choices of imagery. Levy and Barowsky (1986)

indicated that "certain instruments which may require the subject to

choose among possible ansvers predefined by. . .the computer may be

administered vithout altering the subject's response or affecting the

collection of data" (p. 398). According to Levy and Barovsky (1986), a

computerized projective draving technique requires continuous feedback,

tactile, proprioceptive, and visual perceptual integration in order to

simulate the chain of responses provided by direct manipulation of

draving materials.

Summary

Buck's (1948) H-T-P vas developed as a projective test for

measuring intelligence and personality factors. H-T-P vas a forerunner

to the vork of Cohen, Hammer, and Singer (1988). H-T-P serves as the

inspiration of Burns' (1987) K-H-T-P and as an alternative to

Bolander's and Koch's tree tests (Bolander, 1977). Each of the

projective techniques uses tree draving as a stimulus for eliciting

projection of affect. Each of the techniques is based on the belief

that the resulting dravings represent subjects' unconscious vievs of

themselves.

Research (Cassel, et al., 1958; Bieliauskas and Bristov, 1959; and

Bieliauskas and Farragher, 1983) has shovn that tree draving scores on

at least one projective technique, H-T-P, can vary as a result of the

presence of the examiner during the draving phase, the amount of formal

art training possessed by subjects, and the size of the paper used in

the draving phase. The vay draving characteristics are symbolically

perceived by H-T-P subjects may vary depending upon the sex of the

subject (Marzolf and Kirchner, 1972). Research (Bieliauskas and Moens,

1961 and Covden, et al., 1955) has also shovn that H-T-P is more

accurate for making group predictions, but is less accurate for making

individual predictions, depending upon the population of the subjects.

Case studies (Buck, 1948, 1985; Bolander, 1977; Hammer, 1980; Naumberg,

1980; Burns, 1987) have indicated that Individual test results must be

considered in light of the history and past performance of subjects.

Computers have been used successfully in interviev and therapy

capacities (Canoune and Leyhe, 1985; Selmi, et al., 1990). Computer

technology may be used to administer projective techniques that require

subjects to choose among several preselected images vithout altering

subjects' responses or affecting data collection (Levy and Barovsky,

1986). Computer assisted projective techniques that incorporate

draving must duplicate the thought processes used in draving (Levy and

33

Barovsky, 1986).

The next chapter relates how C&T vas developed as a computer

assisted projective technique. In addition, the next chapter presents

the systea used for scoring subjects' tree iaages and the Methods used

for adainistering the coaputer and pencil and paper techniques.

34

CHAPTER III: PROCEDURES

CAT vas designed and administered in order to test three null

hypotheses; that there vas no quantitative difference betveen CAT tree

images and dravn tree images, no quantitative difference betveen

subjects' attitudes regarding the computer generated tree images

spontaneously dravn tree images, and no effect in post-test scores due

to the amount of time betveen tvo administrations of the projective

tree draving technique.

A modified pretest-posttest control group design vas used to test

the null hypotheses. The experimental group vas administered the CAT

projective tree draving technique vhile the control group vas given a

paper and pencil version of the same technique. Both groups vere given

the paper and pencil version as a post-test.

Design Guidelines

Computerizing the projective technique used in this pilot study

involved more than simply creating images based on Buck's H-T-P or

similar techniques and sequencing them on the computer's monitor. CAT

vas created using the folloving four design elements that vere derived

from guidelines for computer-based test design and administration

established by the Committee on Professional Standards and the

Committee on Psychological Tests and Assessment (COPS and CPTA, 1986):

1. CAT administration provided subjects vith at least the same

degree of control and feedback that they experienced vhen taking the

traditionally based pencil and paper version of the tree draving

technique. The computer sequencing process duplicated as closely as

possible the draving process vhile still providing preselected images

35

that served as the basis for objective scoring. A vide variety of

preselected iaages vere aade available to subjects using CAT.

2. The aethod of presentation duplicated as closely as possible

the structure of already existing projective tree draving techniques.

CAT was "user friendly" when presenting stimuli and recording

responses. The design vas easy to use vithout handicapping subjects or

causing them unnecessary frustration.

3. CAT accurately recorded subject responses. The conputer

program vas free of flavs so that data vas not coaproaised. When

subjects aade choices, the results accurately represented those

choices.

4. CAT informed subjects of the perfornance factors that vere

relevant for the pilot study. CAT also aaintalned subject

confidentiality by requiring then to enter a personal code nuaber

before the prograa vould run. Subjects vere free to change their Binds

vhen selecting iaagery or exit froa the prograa vhenever they vished

vithout coaproaising data collection.

By computerizing the projective technique, I took advantage of the

unique flexibility and control of the computer as an art aediua. Based

upon her ovn observations as an art therapist, Canter (1989) found that

therapy subjects vere aore villing to experiaent vith the visual

properties of their vork and take artistic risks vhen using the

aicrocoaputer.

Art teachers (Linehan, 1983; Cleaents, 1985; Greh, 1986; D'Angelo,

1988) vho used the coaputer in instruction reported that students

became actively engaged in the creative process vhen aicrocoaputers and

36

graphics software were used as an expressive medium. Using the

computer, art students easily produced, viewed, and saved a variety of

alternative solutions to a visual problem without the fear of making

mistakes or permanently altering the art piece (Greh, 1986; D'Angelo,

1988). Studies (Schall, Silvestro, & Brown, 1985; Office of Technology

Assessment, 1988) reported that poorly motivated, educationally

disadvantaged, and exceptional learners benefitted more from computer-

assisted instruction than from traditional instructional approaches.

Equipment Used

For the experiment, the computer was set up in an office space

with subdued light to keep the screen free from glare. I created and

presented CAT using the Commodore Amiga 500 computer. The Amiga had

400,000 bytes of chip RAM available for displaying CAT imagery.

Although the Amiga's Motorola 68000 microprocessor and graphics chip

allow the computer to display up to 4,096 colors at once, display of

the CAT program was limited to two colors to conserve chip RAM.

Compute-A-Tree Images

CAT uses a predetermined number of images that subjects selected

to construct their trees. The computer images were created using

Silva's Deluxe Paint II (1986). I used Doyle's The Director (1987) for

image sequencing and text display. In order to create specific tree

images for the menus of CAT, I consulted Buck (1948, 1985), Bolander

(1977), and Burns (1987). In addition, I asked a heterogeneous

population of 136 college students who were not participants in the

experiment and who were non-art majors to create drawings of trees.

From these sources, I formulated the 97 images used in CAT.

37

Buck (1948, 1985) provided illustrations for quantitatively

scoring each of the components of H-T-P in his test manual. The

diagrams only serve to illustrate the text descriptions of Buck's

scoring system; their aesthetic quality is disappointing. The case

study examples in Buck's revised manual are more graphic and visually

pleasing.

Bolander (1977) provided both case study samples of subjects'

works as veil as over one hundred diagrammatic examples of trees and

parts of trees that might be considered when scoring subjects' works.

Bolander (.1977) also interpreted each of the characteristics

illustrated but did not discuss how she developed the interpretations.

Burns (1987) provided case study illustrations and diagrams

illustrating trees and tree features that he indicated as significant.

Burns cited his own research and the research of others when providing

interpretations for specific tree features.

I used the tree drawing techniques of Buck, Bolander, and Burns to

categorize the 136 drawings that were solicited from students. The

catagories discriminated between tree trunks that were one dimensional

(drawn with a single thin or thick line) and trunks that were two

dimensional (drawn with two or more lines). I also used Buck to

classify the ways that students indicated trunk details such as bark,

broken branches, and roots. Student drawings were classified depending

upon how tree branches and foliage were drawn. From these categories,

I developed the computer images that would serve as subject responses

for the experimental group using CAT.

I rendered the CAT tree images in low resolution. According to

Doyle (1987), rendering each of the coaputer images in low resolution

required about 8,000 bytes of memory. By comparison, an image rendered

with 16 colors in high resolution vould require 128,000 bytes allowing,

at most, only three images to be loaded into chip R&M (Doyle, 1987).

Color was considered an extra variable, therefore, images were rendered

in low resolution black and white with a pixel array of 320 wide by 200

high. The black and white display seemed to cause eye strain for pre-

experimental subjects. I changed the display colors from white and

black to pale mauve and dark green in order to make the display more

appealing. The completed computer tree image appeared as dark green on

a pale mauve background rectangle of 12mm by 13.4mm (4 3/4 x 5 1/4

inches). This rectangle served to isolate the subject's image from the

CAT menu and text. CAT menus were displayed using the same color

scheme as tree images. Text was displayed as pale mauve lettering on

the dark green background.

The images were clear and detailed although individual pixels were

noticeable. CAT menu images were about half the size of the subject's

image but still large enough to clearly present details within the

various choices. The letters of the text were 6mm high. Research

(Grabinger, 1983; Steinberg, 1984; and Kearsley, 1986) suggested that

this text size was acceptable. Studies by Steinberg (1984) and

Kearsley (1986) recommended single spacing between lines of type and

double spacing between statements (1984). Reports by Grabinger (1983)

and Kearsley (1986) indicated that attention getting devices in the

text were distracting rather than helpful. I single spaced the text of

CAT and used no attention devices.

39

Tree top and tree bottom images vere separated so that subjects

could elect to "drav" one element before the other. The tree top and

bottom iaages vere double buffered and stenciled to create the

completed tree. Subjects could reselect, change, or redrav both top

and bottom elements using the preselected images. After a subject made

a choice from the menu, the image vas immediately displayed.

Buck's (1985) H-T-P scoring system is specific for deciduous

trees, less so for pine trees, and non-specific for palm trees, saguaro

cacti, potted trees, or trees not typical of North America. When

scoring baseline indications in tree dravings, Buck (1985) indicated

that potted plants and Christmas Trees on vooden bases should receive a

lover score than a horizon line or a horizon line combined vith a short

line that closes the trunk at its base. According to Buck (1985),

branch systems that are indicated by unshaded circular, deltoid, or

oval shapes should be avarded a higher score than linearly dravn branch

systems that lack taper from tvo lines to one. A branch system implied

by a shaded deltoid should receive as high a score as branch systems

indicated vith branch to branch radiation using tapering lines (Buck,

1985).

Even though palm trees are tropical evergreens, they should not be

avarded a score for a branch system if palm frons radiate from the

central point of the trunk (Buck, 1985). When asking about the tree

type in the post-draving interview, Buck (1985) specified that type

means only evergreen or non-evergreen and not specific subspecies.

The pre-experimental survey of 136 college students' tree dravings

shoved a preference for deciduous trees. Slightly over 10% of the 136

40

students surveyed drev evergreen or tropical evergreen trees. One

drawing shoved a pais tree in a pot. Pine trees say have been drawn

because the students were surveyed in November. Pals trees Bay have

been drawn because students attended a southwestern university where

pale trees are plentiful. I designed C&T so that 17 percent of the 57

tree tops available to subjects were evergreen trees (7% pain tree tops

and 10% pine tree tops).

Buck (1985) placed significance in the way the base of the tree is

drawn. In order to sake C&T work successfully, I could only include

the option of indicating or not indicating roots on the tree trunks in

the program. Subjects could not choose between a variety of root

systems for their chosen trunk.

I sequenced C&T images from simple to complex. C&T prompted the

experimental subjects to choose between drawing a tree bottom first or

a tree top. Subjects used a computer mouse to select one of two

labeled icons appearing on the screen. The icons were in the shape of

a seated cat to avoid prompting subjects in the initial selection of

imagery.

Except when making the initial choice of drawing the tree top or

the tree bottom first using the computer mouse, all subsequent input

was through the keyboard. When prompted to do so, subjects selected

images by typing the letter that corresponded to a choice. Each input

key was spaced far enough apart from the others so that subjects would

not accidentally make an undesired selection. No other keys caused the

computer to load images. Pressing an input key automatically entered

subject responses and recorded them on a separate data file. In order

41

to abort the program, subjects had to press the left aouse button.

Operating instructions were provided in the second frame of the program

and on a written card taped next to the key board. Ho prior computer

experience was needed for running Compute-A-Tree or operating the

Amiga.

To initiate the prograa and insure confidentiality, experiaental

subjects typed a personal code nuaber when they were prompted to do so.

Subjects were told to make as good a picture of a tree as they could

using the computer and the provided images. The subjects were also

told that the computer would monitor their selections. They could back

up and change the images as often as they wished and take as long as

they wished. The computer gave a sample of the display area where tree

images would appear and Subjects were again asked if they wanted to

draw a tree top or bottom. After making their choice, aenus were

displayed.

Two sequences were followed depending upon whether subjects chose

to work on the tree top or the tree bottom first. Both sequences used

the same menus of images but these were organized differently. A

facsimile of selected CAT aenus are provided in Appendix C. The

following illustrates the sequence of selections from simple to

complex:

1. Trunk Shape. Subjects chose between a thin single line tree

trunk, a thick single line tree trunk, or a double line tree trunk all

of the same length.

2. Trunk Detail A. Subjects chose to keep their tree shapes or

select a short trunk, a trunk with a broken branch, or the option of

42

reselecting a trunk shape.

3. Trunk Detail B. Subjects who chose a double line tree trunk or

double line tree trunk vith a broken branch vere given this choice.

Subjects vere prompted, "You can add bark by typing A or keep the trunk

shown by typing L". Subjects vho typed A chose betveen two bark

patterns, a random pattern or an "X" pattern or adding no bark.

4. Trunk Detail C. Subjects who chose a tall double line trunk

were also given the choice of adding or not adding a hole to their tree

trunks. The choice of a hole was not dependent on adding bark or a

broken branch.

5. Trunk Roots. Subjects chose between adding or not adding roots

to their trees.

6. Back UP A. Subjects chose between selecting tree top laages

for their trunks or redrawing the tree trunk.

7. Treetop Shape. Subjects chose between a half circle, a

deltoid, an irregular single line with two branches, a circle, and a

single vertical line for their initial tree top shapes.

8. Treetop Detail A. Subjects chose between retaining the shapes

they chose, selecting different shapes, or adding details to the shapes

they chose.

9. Treetop Detail B. Subjects chose between various branch or

foliage shapes aore specific for types of tree tops.

10. Treetop Detail C. Subjects chose between adding leaves or no

leaves froa various patterns. Depending on the top selected, subjects

could also add fruit.

11. Back UP B. Subjects chose to delete details in their tree

43

tops, redraw the entire trees or answer the PDQ questions.

The stimulus for projection by CAT is the computer imagery. The menu

selection and the resulting combinations o£ details renalned finite.

The Pencil and Paper Instrument

As with the computer version, subjects were asked to write in

their personal code number on examination sheets in order to insure

confidentiality. The paper and paper version of the technique elicited

spontaneous tree drawings by asking subjects to make as good a picture

of a tree as they could in a 12mm by 13.4mm rectangle that appeared on

an 8 1/2 by 11 inch page. The size of the rectangle corresponds to the

computer image of CAT.

Subjects were told that they could draw any kind of tree they

wished, that they could erase as much as they wanted, and that they

could take as long as they wished. After completing the drawing,

control group subjects were given a pencil and paper version of the

PDQ. Subjects circled responses on the paper and pencil PDQ. The

questions on both versions of the technique were identical. The pencil

and paper version of the technique was administered as the pretest for

the control group and as the post-test for both the experimental and

control groups.

The Scoring System

Using the scoring systems of Buck (1985), Harris (1963), and

Naglieri (1988) to guide me, I developed a quantitative scoring system

that took into account the various significant tree trunk details, the

branch system, the foliage, and the proportion of the completed

computer image. I used the same scoring system for both computer

isages and pencil and paper drawings as a Beans of comparing the tvo.

Each aajor characteristic received a possible score of four points,

rewarded on the following basis:

1. Trunk. One point was rewarded for a trunk consisting of a

single vertical line or a single line with a broken branch indicated.

Two points were rewarded for indicating trunk diaension with at least a

thick vertical line or silhouette, with or without a broken branch.

Subjects received three points if the trunk consisted of two or nore

vertical lines indicating the outline of a dimensional tree trunk. A

bonus point was given if the trunk consisted of two or aore vertical

lines.

2. Trunk Detail. Subjects received one point if roots were

indicated either actually or by an iaplied taper in the trunk. One

point was given if at least one representation of bark was indicated

including roughness of the outline contour or silhouette, pattern,

shading, single line interior contour line, slash line, or hole.

Subjects received one point if two or aore representations of bark were

coabined. h bonus point was given all trunk details were present.

3. Branch System. One point was given for a branch system that

had actual or iaplied branch to branch radiation. Branch to branch

radiation could be iaplied by drawing a periaeter outline geoaetric or

bioaorphic shape. Pala tree iaages and pine tree iaages that have no

obvious branch to branch radiation receive no credit. One point was

given if the branches were indicated at least by a periaeter outline

with no shading or indicated by wholely single or wholly double lines.

One point was given if the branch system consisted of a shaded

45

periaetec outline, a perimeter outline coabined with single or double

lines, or branches tapering from double lines to single lines.

Subjects received a bonus point if all o£ the branch system details

were indicated.

4. Foliage. One point was given if foliage was shown by leaves,

shading, or designation on the PDQ. One point vas rewarded if

individual leaves or leaves and fruit were indicated, leaves were

indicated by periaeter outline, or a periaeter outline with fruit or a

non-random pattern other than shading. Subjects received one point if

leaves were indicated by shaded periaeter, silhouette, or shaded

perimeter combined with individual leaves. A bonus point was rewarded

if all of the foliage characteristics were present.

5. Proportion. Subjects scored one point if the trunk width at

the base was less than or equal to the width at the branch juncture.

One point was given if the width of the branch system was less than or

equal to the full height of the tree. No point was given if the image

appeared to be chopped by the side of the picture plane as occasionally

occurred in the pencil and paper post test. One point was given if the

trunk, root, and branch dimensions matched at the juncture. No point

was given for a single line trunk that had periaeter outlined branches,

a single line trunk with double line roots or branches, or a double

line or thick single line trunk with thin single line roots or

branches. Subjects received a bonus point if all of the proportion

characteristics were present.

The CAT PDQ consisted of a series of 28 statements and a summative

or Likert scale for subjects' responses. The scale contained a

46

five-category rating system: strongly agree, agree, undecided,

disagree, and strongly disagree (Shav and Wright, 1967). The

corresponding scores of 5, 4, 3, 2, and 1 vere assigned for items

deemed positive and reverse scores vere assigned for items deemed

negative.

CAT Reliability

Research (Cassel, et al., 1958; Bieliauskas and Bristov, 1959; and

Bieliauskas and Farragher, 1983) identified the presence of the

examiner during administration, the prior art training of subjects, and

the size of the drawing paper used to administer the test as three

factors that affected the reliability of H-T-P. I considered each of

these factors as I authored CAT.

Cassel, et al. (1958) discovered that the presence of the examiner

during H-T-P administration had a profound affect on the size and

quality of details in subjects dravings. CAT administration required

that the examiner be present in order to offer subjects technical

assistance when such assistance vas appropriate. According to the COPS

and the CPTA (1986), denying such assistance could have an adverse

effect on test scores for populations lacking computer training.

CAT instructions appeared both in the program itself and on a

vritten card taped near the key board. Directions for the pencil and

paper tree draving technique vere printed on the test itself. The

examiner also presented the instructions orally to both groups of

subjects before testing began. Experimental subjects took CAT in an

office vith subdued light. The office opened into an adjacent office

vhere control group subjects vere administered the pencil and paper

technique. The examiner vas stationed in a chair that provided a viev

of both of the subjects.

Bieliauskas and Bristov (1959) found that subjects with formal art

training scored significantly higher than those that had no art

training. Subjects for this study indicated that they had less than

one year of formal art training. All of the subjects were elementary

teacher candidates enrolled in an art education methods class.

It can not be denied that the subjects received art training

between the pretest and the post-test. The experiment took place

during the second through the fourth veek of the school semester. The

post-test vas administered to subjects during the fifth veek. Some

subjects received training in contour drawing and modeling during the

time span between the pretest and the posttest. No subjects indicated

that they received special training in drawing trees.

The picture planes of both CAT and the pencil and paper

counterpart measured 12mm x 13.4mm. Research by Bieliauskas and

Farragher (1983), found that using a smaller sized picture plane

measuring 4 1/2x5 1/2 inches yielded drawings that were better

proportioned and detailed than those of the standard H-T-P format.

While a smaller sized picture plane did not qualitatively contribute

to CAT computer imagery rendered in low resolution, it vas considered

a desirable factor for the pencil and paper counterpart. Any

factor that contributed to the aesthetic quaility of control

subjects' dravings vas a velcomed balance to the computer images

created by this author and artist!

CAT Validity

48

For some of the PDQ statements, the concept of positiveness or

negativeness was determined by the nature of the experiment. The

theory behind this experiment dictates that subjects projected into the

tree images they created, therefore the statement, "This tree is alive"

was considered positive while "This tree is dead" was considered

negative.

Since the sample population was female, the statement, "If this

tree vere a person, it vould be a girl or woman" was scored positively

to reflect the concept that, had projection occurred, the tree was a

representation of how the subjects viewed themselves. The statement

indicating that the tree was a "boy or man" were scored negatively for

statistical purposes, although the projection of maleness was not in

itself negative. The inclusion of statements regarding the sex of the

tree forced a dichotomy upon the concept when trees may actually be

considered neuter.

In this experiment, the internal consistency of the PDQ was

computed by correlating each response with the total score. This

computation of internal consistency was based upon discussions of the

Likert scale by Shaw and Wright (1967) and Reckase (1984). Five PDQ

statements with correlations below .30 were dropped from consideration

because these statements were either ambiguous or measured variables

inconsistent with the other items of the PDQ instrument.

Dropped from consideration were statements that dealt with the

projected location of the tree. Buck (1985) included the question,

"Where is that tree actually located" in his PDI. In his analysis of

patients' responses to this question, Buck (1985) suggested that the

49

examiner attempt to determine vhat connotation the vords "woods" and

"forest" have for patients who reply that their trees are located in

such places.

The PDQ statement, "I would like to see a tree like this in a

forest", appeared to be too ambiguous in its wording. Two additional

statements, "I would like to see a tree like this in the city" and

"This tree is like one near ay house" also appeared to be too

ambiguous. The three location questions failed to force subjects to

locate their trees. Quite possibly, the questions recorded subjects'

feelings regarding local municipal ordinances that restrict landscaping

for water conservation purposes rather than projections of self image.

Buck (1985) includes statements in his PDI that consider the

facing of the tree and the weather portrayed in the picture. This

experiment's PDQ included two statements regarding the weather: "In

this picture, it is cold and chilly" and "In this picture it is very

hot". Both of these statements were negatively correlated and were

dropped from consideration because they were ambiguous.

Two negatively correlated PDQ statements, "This tree needs care"

and "This tree needs no care" were reclassified as positive and

negative statements respectively. For statistical purposes, ten

positive and eight negative statements in the PDQ were considered. The

PDQ statements appear in the Appendix.

According to Anastasi (1988) a non-verbal test relies on oral

instructions and communication on the part of the examiner and requires

no reading or writing on the part of the subject. By this definition,

H-T-P may be classified as a non-verbal test. A verbal test relies on

50

written instructions and communication on the part of the examiner and

requires reading or writing on the part of the subject (Anastasi,

1988). CAT and the pencil and paper counterpart were designed as

verbal measurements because they relied on written instructions and

required reading on the part of the subjects.

According to Anastasi (1988), it cannot be assumed that similar

verbal and nonverbal tests measure the same constructs. This

observation poses a dilemma when comparing CAT with H-T-P or other

projective tree drawing techniques that are essentially nonverbal. The

constructs measured by CAT may not necessarily be the same constructs

measured by H-T-P or other nonverbal projective techniques. In

comparing the CAT technique with its pencil and paper counterpart, a

similar dilemma is raised; do both the computer and the pencil and

paper versions of the test possess the same number of factors? If the

correlations of the pretest and the posttest are not identical, it can

be assumed that there are different factors involved in the two

versions of the technique.

CAT differs from H-T-P in other respects. Buck's PDI questions

were more open ended than the CAT PDQ statements. Instead of asking

subjects to agree or disagree to the statement, "This is a deciduous

tree (one that looses its leaves)," Buck (1985) asked, "What kind of

tree is that?" and then prompted subjects to identify the type of tree

drawn. H-T-P examinees were not merely asked to agree or disagree to

statements regarding the sex of their trees but prompted to attribute a

gender to their trees through a series of questions (Buck, 1985). The

significance of drawn tree elements became apparent partially through

the dialogue of the PDI. The Likert scale of the CAT PDQ cannot begin

to duplicate that dialogue. The PDQ was designed to provide

quantitative data for comparing experimental and control group

responses to their tree images.

This study makes no claim regarding the cultural fairness of CAT.

A Mexican-American control group subject pointed out that in her

culture, a palm tree, the subject of her drawing, is not only

considered a tree, but a female tree due to the gender of the tree's

name in Spanish. She also drew the tree in a decorative pot and drew

attention to the fact that her's was a tree that would be found inside

a home. She further indicated that although her tree appeared weak and

delicate it was, nevertheless, healthy, suggesting that she understood

that the concepts of health and strength are not necessarily

synonymous.

According to the CAT scoring system, a drawing of a potted palm

tree without the presence of a true trunk or root system would receive

a score of 11 out of a possible 20. Agreeing that the tree was weak on

the PDQ would also result in a low score of 2 out of 5 on that

particular PDQ item. The subject's comments are one indication that

the quantitative scoring of tree imagery and PDQ questions may create

misconceptions about subjects if the scores are rigidly interpreted as

clinical or diagnostic indications of affect. Scoring low on the tree

drawing and PDQ components of this experimental instrument may not

always be indications that projection has not occurred. Some item by

item analysis was required to make a decision regarding the possibility

that projection occurred during each of the observation points of the

52

study.

Summary

The tree drawing projective technique used in this experiment

consisted of two sections. In the first part, subjects created a tree

image. In the second part, subjects responded to PDQ statements

designed to assess their attitudes towards the tree images they made.

The written directions and PDQ statements made this technique a verbal

test. Traditionally, projective techniques that incorporate tree

drawing have been non-verbal.

CAT was designed to automate administration of the traditionally

based projective technique used in this experiment. CAT allowed

subjects to create tree images using 97 preselected tree top and tree

bottom images. These images were displayed in low resolution in two

colors. CAT kept a record of subjects choices in the tree drawing

component of the technique. CAT also presented the PDQ statements and

automatically recorded their responses.

53

CHAPTER IV: RESULTS

The first and second null hypotheses assert that equivalency

exists between computer administration and conventional administration

of the projective tree drawing technique. If the computer and the

conventional versions of CAT are equivalent, they must have similar or

identical effects, values, and meanings.

According to the guidelines for computer-based testing established

by the COPS and the CPTA (1966), computer versions of tests may be used

in place of conventional tests when scores for the two versions are

equivalent and when computer specific factors have no appreciable

influence on scores. There are two prerequisites to equivalency (COPS

and CPTA, 1986):

1. The rank orders of scores of individuals taking both forms of

the test must approximate each other.

2. The scores of both versions have approximately the same means,

dispersions and distributions.

If the first prerequisite is met but not the second, the computer

scores can be rescaled so that they are comparable. If the second

prerequisite is met but not the first, the tests may be distributed in

the same way but not be equivalent. In other words, the two tests are

not measuring the same constructs and cannot be used interchangably if

the first prerequisite is not met (COPS and CPTA, 1986).

In this chapter, CAT results are analyzed to assess the

similarities and differences that were observed in pretest and posttest

scores for the experimental and control groups. £ tests were used to

compare means within and across groups. Computer and pencil and paper

54

scores for the experiaental group vere put in rank order. Correlations

were used to identify associations. Tables of data appear in Appendix

E.

Table 1 provides a statistical description of the results of the

computer administered and pencil and paper administered versions of the

projective tree drawing technique. The means for the total scores of

the control group shifted slightly upvard from the pretest to the

posttest with only a 1.94 difference. The means for the total scores

of the experimental group shifted dovnvard from CAT to the posttest

with a 3.37 difference. The difference between the standard deviations

for the control group test was +.05 while the difference for the

experimental group tests was -.81.

Table 2 lists the rank order of pretest and posttest tree and PDQ

scores for the experimental group, the only group taking both the

computer version and the conventional version of the projective tree

drawing technique. The rank order of individual scores is not

consistent for the two versions of the technique for experimental

subjects.

Correlations

Table 3 provides the correlation coefficients for pretest and

posttest tree scores and PDQ scores for the experimental and control

groups. There was only a moderate association between pretest and

posttest tree scores for the experimental group (r = .183). There

appeared to be an association between pretest and posttest tree scores

for the control group (r = .376). The correlation of control group

tree drawing scores was significant with alpha = .05 while the

55

correlation of experimental group tree scores failed to reach

significance. Only a moderate association (r = .279) appeared between

pretest and posttest PDQ scores for the experimental group. A strong

(r = .835) relationship appeared between pretest and posttest PDQs for

the control group. Negative correlations occurred for the pretest tree

scores and the pretest PDQ scores for both groups while moderate

correlations occurred between posttest tree drawing scores and PDQs.

Variance Within Groups

A paired £. test was computed to see if the mean of the difference

between the pretests and posttests for the sample groups was different

from zero. Table 4 shows the £ ratio and p values for the first set

of observations and the posttest scores of the experimental and control

groups.

The mean of the difference between pretest and posttest tree image

scores for the experimental sample is not significantly different from

the population with alpha = .05. The mean of the difference between

pretest and posttest tree image scores for the control sample is also

not significant. The mean of the difference for the experimental

group's PDQ scores is not significant (p = .200). In contrast, the

mean of the difference for the control group's PDQs is significant

(p = .05).

Variance Across Groups

Using an independent t test, an across group test of variance was

made. The results appear in Table 5. The degrees of freedom were

computed by an approximation (Ryan, Joiner, and Ryan, 1985). With

alpha = .05, the independent £ test shows that there is a significant

56

difference betveen means for pretest experimental and control group

tree images (p = .026). The difference betveen the neans for posttest

experiaental and control group tree iaages is not significant

(p = .65). There appeared to be no difference betveen the means of

experiaental and control group pretest PDQ scores. Significance vas

not found betveen posttest PDQ means for experimental and control

groups.

The Effect of Time

Subjects vere timed during the tree draving component of the first

set of observations. Experimental subjects did not save any time by

constructing their tree using the computer menu. Experimental subjects

took, on the average, 8.9 minutes to construct a tree on the computer.

Control subjects took, on the average, 8.8 minutes to drav their trees.

The number of days betveen the pretest and the posttest varied

from subject to subject vithin both groups. The time span betveen the

pretest and the posttest vas recorded for each subject. A tvo-vay

analysis of variance (ANOVA) vas computed to see if posttest scores

vere affected by the number of days betveen the tvo tests. The first

independent variable for the ANOVA vas group association. The second

independent variable vas the number of days betveen administration of

the pretest and posttest. The data vere unbalanced because of the

number of subjects in each group and because different numbers of

subjects came in during each of the days of the experiment.

The results appear in Table 6. No main effects for either

independent variable and no interactions are evident for posttest tree

dravings. ANOVA results shov no main effect for the first independent

variable, a Bain effect for the second independent variable, and an

interaction for posttest PDQs. Figure 1 appears in Appendix 1. Figure

1 graphs the posttest scores for the tvo groups by the number of days.

Table 6 provides the Bean scores for the tvo groups by the nuaber of

days. Interaction appears to be evident fifteen days froa the pretest

to the posttest.

Specific PDQ Stateaents

Given the PDQ data provided by this study and the suaaative

lnstruaent used to obtain the data, there is no concrete way of telling

if subjects projected into the tree images they created. But answers

to specific questions Bight suggest that projection took place. Table

7 provides the percent of subjects answering each of the five responses

to five PDQ questions. Correlations for each of the five questions and

the total score are also included vith Table 7. In addition,

correlations of each PDQ question and the total PDQ score are included

vith the PDQ provided in Appendix B.

I found that 48% of experiaental subjects said that their coaputer

tree laages reainded thea of a girl or voaan, however an equal percent

said that their coaputer iaage did not realnd thea of a girl or voaan.

Of those that sexed their tree as feaale, 7% also said that their tree

vas aale. A total of 20% aarked "disagree" or "strongly disagree" on

both of the sex questions.

The correlation of responses to the stateaent, "If this tree vere

a person, it would be a boy or aan" vith the PDQ total score was

neither significant in the pretest nor the posttest for the

experiaental group. There vas high association between the statement,

58

"If this tree vere a person, it would be a girl or voman" and the

pretest PDQ score (r = .413) There vas also high association between

the statement and the posttest PDQ score (r = .471). Both correlations

were significant at alpha = .05, suggesting that there is less than 5

chances in 100 that the correlation is significant by chance alone.

During the posttest, 44% of the experimental subjects marked

"disagree" or "strongly disagree" on the tree sex question. The number

of experimental subjects who responded by disagreeing to the tree sex

statements in the posttest almost doubled from that of the pretest.

1 found that 15% of the control group subjects were undecided when

asked to sex their first tree drawings. A total of 52% of the control

subjects sexed the pretest tree as female. Of those that agreed to sex

their trees as females, 7% also agreed to sex their trees as males.

During the posttest, 36% of the subjects disagreed to sex their trees

while 3% sexed the tree as both male and female.

Correlation of responses to the statement, "If this tree were a

person, it would be a girl or woman" reached significance (r = .483).

Posttest correlation of responses to the same statement and total score

also reached significance (r = .392).

The computer image reminded 19% of the experimental subjects of

someone they knew while 22% of the experimental group said that their

posttest drawn tree reminded them of someone they knew. For the

experimental group, there was low association between the statement

and the total PDQ score on the computer version (r = .172). This

correlation failed to reach significance indicating that the

relationship may be due to sampling error. There was higher and

59

significant association between the statement, "My tree reminds me of a

person I knov" and the total PDQ score for the experimental group in

the conventionally administered posttest (r = .446).

Control group subjects answered consistently on the associative

question. The same 26% of the control subjects who said their tree

reminded them of a person during the pretest also answered

affirmatively when presented with the concept in the posttest. For the

control group, the correlation of pretest responses to the statement,

"My tree reminds me of a person I know" with pretest total PDQ scores

reached significance (r = .411). A correlation of control subjects'

responses to the same statement with posttest scores also reached

significance (r = .424). No subjects in either group were undecided

when presented with the associative statement.

Five experimental subjects found making the computer tree

difficult. &11 five agreed that making a tree drawing was not

difficult. Four experimental subjects found making the computer tree

easy while they agreed that making a tree drawing was hard.

Association between the statement, "I found making a picture of a tree

difficult" and total PDQ scores for the experimental group was high

(r = .558) on the pretest and low (r = .283) on the posttest. The

pretest correlation reached significance while the posttest correlation

did not, indicating that posttest responses to the statement by the

experimental group were not above chance.

Fewer control subjects found making a tree drawing difficult

during the posttest than during the pretest. Pretest correlation

(z = .471) and posttest correlation (r = .627) reached significance for

60

the control group.

A total of 81% of the experimental group liked the computer images

they created using CAT while 68% of the control group liked their first

drawing. A total of 89% of the experimental group liked their post-

test drawings while 80% of the control group liked theirs. Those

experimental subjects who complained usually said they wanted to

explore the CAT program more by making a second tree. Correlations

between experimental group responses to the statement, "I like the way

my tree looks" and total PDQ scores for the pretest (r = .299) and the

posttest (r = .313) did not reach significance in either case.

A greater number of control subjects liked their posttest drawing

than those who liked their pretest drawing. Fewer control subjects

found drawing a tree difficult the second time around. Correlations

for pretest and posttest responses to the statements, "I found making a

picture of a tree difficult" and "I like the way my tree looks" reached

significance at alpha = .05.

Summary

Correlations of experimental group tree and PDQ scores were

positive but failed to reach significance. Correlations of control

group tree and PDQ scores were positive and significant.

Paired £ tests of the means of the differences between

experimental posttest and pretest trees and PDQs yielded results that

were not significant. The same paired t tests yielded significant

results for control group posttest and pretest PDQs.

An independent t test made across group comparisons of the pretest

and posttest trees and PDQs for the two sample populations. The means

61

of pretest trees for the two samples were significantly different.

Independent £. tests of the means of pretest PDQs, posttest trees, and

posttest PDQs failed to shov significance.

A two way ANOVA of variance was computed with posttest tree and

PDQ scores as the dependant variable and group association and days

between the pretest and posttest as independent variables. The results

showed that no main effects or interaction of either independent

variable occurred with tree drawings, but yielded a main effect for the

independent variable of days between treatments and an interaction for

posttest PDQ scores.

The next chapter discusses the results of the experiment. The

thesis questions are answered and recommendations are made for future

investigations of computer assisted projective techniques.

62

CHAPTER V: DISCUSSION

Three null hypotheses vere investigated through this study. Tests

of the data vere conducted vith alpha = .05. The first null hypothesis

states:

HOI There vill be no quantitative difference between CAT tree

images and dravn tree images.

The evidence supports rejection of HOI. The independent t test

found that the difference between the means of CAT image scores and

pretest control group tree drawing scores was significant. The CAT

mean score was 1.85 points higher than the mean score of the pencil and

paper counterpart. The same independent £ test revealed no significant

difference between the mean scores for experimental and control group

posttest tree drawings. The paired £ test revealed that the difference

between the means of experimental posttest and pretest tree drawings

was not significant with alpha = .05. The paired £ test also showed

that the difference between the means of control group posttest and

pretest tree drawings was not significant. The two way ANOVA found no

effect by the independent variable of group association upon posttest

tree drawing scores.

The second null hypothesis states:

H02 There will be no quantitative difference between subjects'

attitudes regarding the computer generated tree images and subjects'

attitudes regarding the spontaneously drawn tree images.

There is not enough evidence to reject H02. Correlations of the

experimental pretest and posttest PDQ scores were positive but not

significant. In contrast, the correlation between control group

63

pretest and posttest PDQs was both positive and highly significant.

The paired £ test of the experimental group revealed that the

difference between posttest and pretest means of the PDQs was not

significant. However, the paired £ test of the control group revealed

that the difference between the means of the PDQs was significant. The

independent £ test found that there was no significant difference

between the pretest PDQ means for the two groups. The independent £

test also found that there was no significant difference in posttest

PDQ means for the two groups. The two way ANOVA demonstrated no effect

by the independent variable of group association upon posttest PDQ

scores.

The sample size of the experimental group (N = 27) may be too

small to accurately consider the question of equivalency. In the case

of CAT'S tree drawing component, equivalency with the conventional

counterpart cannot be claimed because rank orders of scores changed

between computer administered and conventionally administered tests for

the experimental group. In addition, the means between computer and

conventional pretests were significantly different. In the case of

CAT's PDQ component, rank orders also changed between computer

administration and conventional administration, but I believe there is

insufficient evidence to answer the question of equivalency between the

two forms of administration.

The third null hypothesis states:

H03 Post-test scores will not be influenced by the amount of time

between two administrations of the projective tree drawing technique.

The two way ANOVA demonstrated a main effect for the independent

64

variable o£ the number of days between test administrations and

interaction between that variable and posttest PDQ scores. The

relationship between the two variables appears to represent a cubic

trend rather than a linear trend. The demonstrated interaction between

days between pretest and posttest administration and posttest PDQ

scores justifies rejection of H03.

In addition to testing each of the null hypotheses, three thesis

questions were asked:

1. Did subjects express favorable attitudes about the trees they

create on the computer using preselected imagery? Was that attitude

different from subjects who drew trees?

2. Did subjects' responses on the CAT PDQ tend to indicate that

they relate to the computer images in the same way that they relate to

spontaneously drawn images?

3. Did the form of computer administration used in this

experiment, the decision-inducing set (DIS), influence subjects'

responses?

Subjects' responses to specific PDQ questions suggest that the

computer tree image was less difficult to produce than the tree

drawing. Specific responses also suggest that subjects were less

inclined to personal associations with preselected computer imagery

than with spontaneous original drawings. Although PDQ responses might

suggest that subjects' attitudes toward CAT trees were different from

drawn trees, further study would be necessary to answer the first and

second thesis questions.

Pretest means obtained through computer administration were higher

65

than conventionally obtained pretest means. The menu and prompts might

have influenced some subjects to change their tree images, but the

change did not necessarily lead to a higher tree score. Of the 12

experimental subjects who chose to redraw their entire CAT tree image

during computer administration, five subjects produced a second image

that scored lower than their first tree. Five experimental subjects

who elected to redraw their CAT trees produced images that received

identical scores. Two subjects who chose to completely redraw their

CAT tree image produced a second image that scored higher than their

first image.

Two experimental subjects indicated that they included fruit in

their CAT image because the prompt asked them if fruit was a detail

they desired to add. Although the prompt may have suggested the

addition, the DIS that followed allowed subjects to remove the fruit.

There is no evidence that the DIS caused the higher mean score in the

pretest experimental group. There is also no evidence that computer

administration of the pretest affected posttest scores for the

experimental group.

Conclusions

Further analysis is necessary to identify the factors that

contributed to an appropriate and valid PDQ. Future research should

focus on the validity of the tree drawings. Incorporating more tree

images and decision-making sets would greatly enhance CAT as an

alternative to H-T-P and other conventional techniques.

CAT has several advantages over conventional administration of the

tree drawing technique. The technique could be administered to special

66

and handicapped populations. The administration and scoring would be

easier and more objective. Scoring would be consistent and reliable.

Variables reflecting the differences in examiners would be decreased,

if not eliminated by computer administration.

Computer technology is advanced enough that computer prompts could

be provided and an open ended PDQ could be administered through speech

synthesis linked with the projective program. Computer administration

could closely approximate the nonverbal format of traditional

projective tree drawing techniques. Through this approach, subjects

may feel freer to indicate what they are actually projecting about

their trees.

67

APPENDIX A

CAT QUANTITATIVE SCORING SYSTEM

1. Trunk. Score one point for each characteristic included.

A. At least a trunk consisting of a single vertical line or line vith

broken branches.

B. An indication of trunk dimensional rendering through the use of at

least a thick vertical line (or silhouette) vith or without broken

branches.

C. The Inclusion of two or more vertical lines indicating the outline

or contour of a dimensional tree trunk.

D. Score one bonus point if both the dimension and contour are

included. Maximum total: 4 points.

2. Trunk Detail. Score one point for each characteristic

included.

A. Roots present by either actual indication or implied taper of the

trunk base.

B. At least any one representation of bark (including roughness of a

silhouette trunk, a pattern, shading, a single interior contour line or

slash line or hole).

C. Two or more representations of bark combined.

D. Score one bonus point if all of the above are present. Maximum

total: 4 points.

3. Branch System. Score one for each characteristic.

A. Branch system has actual or implied (as vith a perimeter outline)

branch to branch radiation (exceptions: pine or palm trees and cactus

without branch to branch radiation).

B. One of the following:

1. a perimeter outline of branches with no shading.

2. vholely single line branches.

3. vholely outlined (2-D) branches.

C. One or more of the folloving:

1. a perimeter outline combined with vholely single or outlined

branches.

2. branches tapering from double lines to single lines.

3. a shaded perimeter outline.

D Score one bonus point if A-C are present. Maximum total: 4

points.

4. Foliage. Score one for each characteristic present.

A. Foliage shovn by leaves, shading, or designation on the PDQ.

B. One of the folloving characteristics present:

1. individual leaves indicated or leaves and fruit.

2. perimeter outline.

3. perimeter outline with fruit or pattern other than shading.

C. Leaves indicated by shaded perimeter (cloud shape), silhouette, or

shaded perimeter combined vith individual leaves.

D. Score one bonus point if all of the above characteristics are

present. Maximum total: 4 points.

5. Proportion. Score one for each characteristic included.

A. Trunk vidth at base greater than or equal to width at branch

juncture.

B. Width of branch system less than or equal to full height (vhen the

image is paper chopped on three sides, score 0).

69

C. Trunk dimension branch dimension and root dimension match at

junctures (score 0 for single line trunk with perimeter outline

branches, single line trunk vith double line branches or roots, and

double line trunks or thick single line trunks vith single line roots

or branches).

D. Score one bonus point if all characteristics are included.

Maximum total: 4 points.

70

APPENDIX B

POST DRAWING QUESTIONNAIRE (PDQ)

With Experimental Pretest and Posttest Correlations (01 and 02)

And Control Pretest and Posttest Correlations (03 and 04)

The following statements appeared in pretest and posttest PDQs

fpr both groups. Subjects responded on a five point summative scale

with strongly agree, agree, disagree, or strongly disagree. No opinion

vas indicated by circling all of the answers or typing return on the

computer. Correlations with total scores appear below each statement.

This is a deciduous tree (one that looses its leaves).

01 02 03 04

.383 .138 .074 .177

If this tree were a person, it would be a boy or man.

01 02 03 04

.162 .333 .290 .208

If this tree were a person, it would be a girl or woman.

01 02 03 04

.502 .471 .483 .392

This tree is dead.

01 02 03 04

.495 .391 .300 .334

This tree is lonely.

01 02 03 04

.376 -.065 .486 .405

This tree is part of a group.

01 02 03 04

.152 -.094 .480 .200

My tree reminds me of a person I know.

01 02 03 04

.181 .446 .411 .424

My tree is weak.

01 02 03 04

.232 .392 .314 .284

My tree is healthy.

01 02 03 04

.454 .506 .562 .644

This tree needs care.

01 02 03 04

.623 .674 .255 .413

My tree is strong.

01 02 03 04

.467 .389 .238 .406

My tree is old.

01 02 03 04

.220 .373 .527 .371

My tree is young.

01 02 03 04

.244 .440 .540 .255

This tcee is alive.

01 02 03

.544 .414 .281

This tree is sickly.

01 02 03

.498 .424 .371

This tree needs no care.

01 02 03

.626 .558 .512

1 found making a picture of a

01 02 03

.571 .283 .471

I like the vay my tree looks.

01 02 03

.303 .313 .595

04

.322

04

.454

04

.383

tree difficult.

04

.627

04

.531

73

APPENDIX C

FIGURE 1: FACSIMILES OF SELECTED CAT MENUS

The Coapute-A-Tree aenu reproduced belov Is approxlaately one half

the actual size. Because of reduction and printer translation,

reproduced inages and text lack the fidelity of the original. Prior to

the decision-inducing set (DIS) shown, a two dinensional tree trunk was

selected with bark, roots, and broken branch details. When subjects

chose A, a menu of drooping tree tops appeared that included willow and

palm trees. When subjects chose D, pine trees appeared. Choices G and

J displayed deciduous tree tops while choice L displayed tree stumps

and a telephone pole like tree.

Choose a tree top shajpe A by typing the letter next

to t h e p i c t u r e . Y o u wi l l be able to add details to these shapes to Make

p different kinds of trees.

G

J

L

74

FIGURE 1: Continued

Subjects vho selected G in the previous DIS were given the choice

of adding one of a variety of branch to branch systems including

vholely single line branches, vholely outlined branches, and branches

that tapered from double lines to single lines. After selecting a

branch system, subjects vere given the option of adding leaves. Those

subjects vho chose to add leaves vere presented vith the DIS shovn

belov. The silhouette tops shovn vere scored as double line branches

that tapered from double to single lines.

A

D

Choose A, T>, or G or Type: Z for no leaves

G M to see nore leaves

75

APPENDIX D

HOUSE-TREE-PERSON POST DRAWING INTERROGATION (PDI)

(For Tree Drawings)

Buck's (1948, 1985) PDI is provided for comparison with the CAT PDQ.

The following questions are presented in the order they appear in

The House-Tree-Person Technique; Revised Manual (Buck, 1985).

Explanations of the questions and the H-T-P scoring systea aay be found

by consulting the sanual.

1. What kind of tree is that?

a. Is it the sort of tree that stays green the year 'round, or

does it drop its leaves?

2. Where is that tree actually located?

3. Hov old is that tree?

4. Is that tree alive?

5. a. What is there about the tree that gives you the iapression

that it is alive?

b. Is any part of the tree dead?

c. What do you think caused it to die?

d. Hov long has it been dead?

6. Which does that tree look aore like to you: a Ban or a voaan?

a. You've probably seen rugged, powerful, robust trees that Bade

you think of a Ban, and other trees that looked either graceful

and tria as a young voaan or as large, protective, and aotherly

as an older voaan. Which does this sake you think of, a Ban or

voaan?

b. Does any part of this tree Bake you think of a Ban or a voaan?

76

7. ihat is there about the tree that gives you that inpression?

8. If that vere a person instead of a tree, vhich vay would the

person be facing?

9. Is that tree by itself, or is it in a group of trees?

a. Do you think it would like to be in a group?

10. As you look at that tree, do you get the iepression that it is

above you, belov you, or about on level vith you?

11. What is the weather like in this picture?

12. Is there any wind blowing in this picture?

13. Show ne the direction it is blowing?

14. What sort of wind is it?

a. How would you feel about such a wind?

Standard Deviation Standard Error of Mean

APPENDIX E

D&T& ANALYSIS

Table 1.

Description of Pretests and Posttests for Experimental and Control

Groups

Experimental Group N = 27

Pretest (CAT)

Mean Median

Tree 16.37 16.00 3.07 .585

PDQ 66.67 67.00 7.57 1.46

Posttest

Mean Median

Tree 15.11 15.00

PDQ 64.56 65.00

Control Group N = 31

Pretest

Mean Median

Tree 14.52 14.00

PDQ 64.10 63.00

Posttest

Mean Median

Tree 14.74 15.00

PDQ 65.81 65.00

Standard Deviation Standard Error o£ Mean

2.83

6.15

.545

1.18

Standard Deviation Standard Error of Mean

3.129

8.40

.562

1.51

Standard Deviation Standard Error of Mean

3.29

7.59

.590

1.36

Table 2.

Rank Order of Pretest and Posttest Scores for the Experimental Group

<N = 27)

Subject Pretest (Computer) Rank Posttest Rank

Tree PDQ Tree PDQ

01 8.12 11.5 4.5 8.33

04 8.12 3.33 3.33 2.5

08 7.33 13.0 5.2 2.5

09 1.0 1.3 1.25 5.5

11 8.12 3.33 1.25 12.5

13 5.33 7.5 3.33 3.5

15 5.33 5.5 5.2 1.0

17 5.33 14.33 6.0 4.5

19 6.2 8.0 5.2 3.5

22 8.12 3.33 1.25 4.5

62 3.25 15.0 1.25 15.0

63 6.2 11.5 3.33 8.5

65 6.2 6.0 7.5 7.33

66 8.12 9.5 8.16 3.5

69 6.2 2.0 9.5 7.33

71 6.2 12.0 7.5 13.0

74 4.0 1.33 2.5 6.0

75 8.12 10.0 8.16 16.0

76 3.25 14.33 8.16 14.0

81 2.0 16.5 4.5 9.0

83 3.25 7.5 8.16 12.5

Table 2 Continued.

Subject Pretest (Computer) Rank Posttest Rank

Tree PDQ Tree PDQ

86 6.2 14.33 2.5 11.5

87 8.12 16.5 5.2 8.33

90 8.12 4.0 9.5 7.33

91 3.25 5.5 5.2 10.5

94 7.33 9.5 8.16 11.5

98 7.33 1.33 8.16 10.5

Table 3.

Correlations of pretest and posttest scores

Experimental Group N = 27

CAT Tree

CAT PDQ -.093

Posttest Tree .183

Posttest PDQ -.162

Control Group N = 31

Pretest Tree Pretest PDQ Posttest Tree

Pretest PDQ -.102

Posttest Tree .376 -.067

Posttest PDQ -.017 .835 .148

CAT PDQ Posttest Tree

.018

.279 .122

80

Table 4.

Paired T-Test (Two Tailed) of the Difference between Posttest and

Pretest Scores

Test of MU = 0 vs MU N. E. 0

Experimental Group N = 27

Catagory Mean Standard Deviation

Tree -1.26 3.76

PDQ -2.11 8.31

Degrees of Freedom T

Tree 26 -1.74

PDQ 26 -1.32

Control Group N = 31

Catagory Mean Standard Deviation

Tree .226 3.58

PDQ 1.71 4.66

Degrees of Freedom T

Tree 30 .35

PDQ 30 2.04

Standard Error of Mean

.723

1.60

P value

.093

. 2 0 0

Standard Error of Mean

.644

.836

P value

.730

.050

SI

Table 5.

Two Sanple Independent T Test of Experimental and Control Group Scores

Pretest Tree Scores

N Mean Standard Deviation Standard Error of Mean

Experimental 27 16.37 3.04 .585

Control 31 14.52 3.13 .562

Confidence Interval (95%) T P Degrees of Freedom

(.2281, 3.480) 2.29 .026 55

Pretest PDQ Scores

N Mean Standard Deviation Standard Error of Mean

Experimental 27 66.67 7.57 1.46

Control 31 64.10 8.40 1.51

Confidence Interval (95%) T P Degrees of Freedom

(-1.634, 6.774) 1.23 .23 55

Posttest Tree Scores

N Mean Standard Deviation Standard Error of Mean

Experimental 27 15.11 2.83 .545

Control 31 14.74 3.29 .590

Confidence Interval (95%) T P Degrees of Freedom

(-1.241, 1.980) .46 .65 55

Posttest PDQ Scores

N Mean Standard Deviation Standard Error of Mean

Experimental 27 64.56 6.15 1.18

Control 31 65.81 7.59 1.36

Confidence Interval (95%) T P Degrees of Freedom

(-4.869, 2.368) -.69 .49 55

Table 6.

Analysis of Variance Showing Degrees of Freedom (DF). Suns of Squares

(SS). Mean Squares (MS). F-Ratio (F). and P Value (P)

Analysis of Variance: Posttest Tree Draving Scores (N = 58)

Source DF SS MS F P

Group 1 2.412 2 .412 0.232 0 .633

Days 9 89.692 9 .966 0.960 0 .487

Interaction 9 41.982 4 .665 0.449 0 .899

Error 38 394.467 10 .381

Multiple R: .512 Squared Multiple R: .262

Analysis of Variance: Posttest PDQ Scores (N ' = 58)

Source DF SS MS F P

Group 1 6.197 6 .197 0.204 0 .654

Days 9 738.692 82 .077 2.699 0 .016

Interaction 9 717.475 79 .719 2.621 0 .018

Error 38 1155.717 30 .414

Multiple R: .736 Squared Multiple R: .542

Table 7.

Means of Posttest PDQs

Days 4 8 9 10 11 14

Experiaental 0.00 58. 00 61.25 58.50 60.00 66.20

Control 61.00 71. 33 60.50 69.00 66.66 67.25

Days 15 16 17 18 20

Experiaental 69.00 72. 00 63.75 64.00 70.50

Control 74.75 59. 00 59.00 48.00 64.00

Figure 1

Main Effect for Davs Only and an Interaction

PDQ Score

90 +

60 +o o * *

o2 *o *2o

o o2 *

*o3 «2o *

*o3

o o

*2o2

o *2 o *

*

0

*0

40

30 + I I I I I I I I I I I 4 8 9 10 11 14 15 16 17 18 20

Munber of Days Between Tests

Control Group = o Experimental Group = *

84

Table 8.

Percent of Subjects Responding to PDQ Statements

If this tree vere a person, it vould be a girl or woman.

Experimental Group

Pretest (r = 0.413)

Posttest (r = 0.471)

Control Group

Pretest (r = 0.483)

Posttest (r = 0.392)

SA

22

11

16

13

A

26

22

19

29

U

4

4

10

3

SA A U

Experimental Group

Pretest (r = 0.184) 4 22

Posttest (r = 0.333) 7 11

Control Group

Pretest (r = 0.290) 10 39

Posttest (r = 0.208) 13 13

My tree reminds me of a person I know.

SA A

Experimental Group

Pretest (r = 0.172) 4 15

Posttest (r = 0.446) 4 18

Control Group

Pretest (r = 0.411) 7 19

Posttest (r = 0.424) 7 19

41

48

32

29

If this tree vere a person, it vould be a boy or man.

41

63

26

39

48

67

58

58

SD

7

15

23

26

SD

29

15

19

35

SD

33

11

16

16

Table 8 Continued.

I like the vay ay tree looks.

SA A

Experiaental Group

Pretest (r = .299) 37 44

Posttest (r = .313) 15 74

Control Group

Pretest (r = .595-) 10 58

Posttest (r = .531) 19 61

I found making a picture of a tree difficult.

SA A

Experimental Group

Pretest (r = .558) 7 7

Posttest (r = .283) 0 19

Control Group

Pretest (r = .471) 16 13

Posttest (r = .627) 10 7

REFERENCES

American Psychiatric Association (1987). Diagnostic and statistical

manual of aental disorders (3rd ed. rev.}. Washington, D.C.:

American Psychiatric Association.

Anastasi, A. (1988). Psychological testing (6th ed.). New York:

Hacmillan.

Bieliauskas, V. J., & Bristov, R. B. (1959). The effect of formal

art training upon the quantitative scores of the H-T-P. Journal

of Clinical Psychology. 15.(1), 57-59.

Bieliauskas, V. J., & Farragher, J. (1983). The effect of change

of the size of the drawing sheet on the H-T-P IQ scores. Journal

of Clinical Psychology. 39.(6), 1033-1034.

Bieliauskas, V. J., & Moens, J. F. (1961). An investigation of the

validity of the H-T-P as an intelligence test for children.

Journal of Clinical Psychology. 11(2), 178-180.

Bolander, K. (1977). Assessing personality through tree drawings.

New York: Basic Books.

Brown, D. T. (1984). Automated assessment systems in school and

clinical psychology: Present status and future directions.

School Psychology Review. 13.(4), 455-460.

Buck, J. N. (1948). The H-T-P Technique: A qualitative and

quantitative scoring manual. Journal of Clinical Psychology.

1(4), 319-396.

Buck, J. N. (1985). The House-Tree-Person technique: Revised

manual. Los Angeles: Western Psychological Services.

Burns, R. C. (1987). Kinetic-House-Tree-Person drawings: An

interpretive manual. Nev York: Brunner/Hazel.

Campbell, D. T., & Stanley, J. C. (1963). Experimental and

quasi-experimental designs tor research. Boston: Houghton

Mifflin.

Canoune, H. L., & Leyhe, B. V. (1985). Hunan versus computer

interviewing. Journal of Personality Assessment. 12.(1),

103-106.

Canter, D. S. (1989). Art therapy and computers. In H. Vadeson,

J. Durkin, & D. Perach (Eds.). Myanma In ftKt ThffMPy. Hev

York: John Wiley and Sons.

Cassel, R. H., "Johnson, A. P., & Burns, 9. H. (1958). Examiner,

ego defense, and the H-T-P test. Journal of Clinical Psychology.

11(2), 157-160.

Clements, R. D. (1985). Adolescents' computer art. Art Education.

2i(2), 6-9.

Cohen, B. M., Hammer, J. S., & Singer S. (1988). The diagnostic

drawing series: A systematic approach to art therapy evaluation

and research. The Arts in Psychotherapy. 11(1), 11-21.

Committee on Professional Standards and Committee on Psychological

Tests and Assessment. (1986). Guidelines for computer-based

tests and interpretations. Washington, D.C.: American

Psychological Association.

Cowden, R, C., Deabler, H. L., & Peamster, J. H. (1955). The

prognostic value of the Bender-Gestalt, H-T-P, TAT, and Sentence

Completion Test. Journal of Clinical Psychology. 11(3), 271-275.

D'Angelo, J. (1988). Computers for art teachers. Art Education. 41

(5), 41-44, 47-48.

Doyle, K. (1987). The director [Computer program]. Torrance, CA: Right

Ansvers.

Grabinger, R. S. (1983). Legibility and computer-generated text.

In M. S.Ford (Ed.). Conference Proceedings: Microcomputers in

education: Forward to the 3 C's: Communicating, calculating.

and computing. Tempe: Arizona State University.

Greh, D. (1986, January). Technology and art. Electronic Learning,

pp. 29-32.

Hammer, E. F. (1968). Projective drawings. In A. I. Rabin (Ed.),

Projective techniques in personality assessment (pp. 366-393).

Hammer, E. F. (1980). The clinical application of projective

drawings (6th ed.). Springfield, IL: Charles C. Thomas.

Hammer, E. F. (1988). Book Review: Kinetic-House-Tree-Person

Drawings: An Interpretive Manual. The Arts in Psychotherapy.

15(2), 165-166.

Harris, D. B. (1963). Goodenough-Harris Drawing Test manual.

New York: Harcourt Brace Jovanovich.

Kearsley, G. P. (1986). Authoring: A guide to the design of

instructional software. Reading, MA: Addison-Wesley.

Killian, G. A. (1984). House-Tree-Person technique. In D. J.

Keyset & R. C. Sweetland (Eds.), Test critiques (Vol. 1,

pp. 338-353). Kansas City, MO: Test Corporation of America.

Levy, A. J., & Barowsky, E. I. (1986). Comparison of computer

administered Harris-Goodenough Drav-A-Man Test with standard

paper and pencil administration. Perceptual and Motor Skills.

£2(2), 395-398.

Linehan, T. E. (1983). Computer graphics: Opportunity for artistic

vision. Art Education. 26(3), 11-14.

Marzolf, S. S., & Kirchner, J. H. (1972). House-Tree-Person

drawings and personality traits. Journal of Personality

Assessment. ££(2), 148-165.

Miller, F. E. (1986). The development and evaluation of an online

computer-assisted Rorschach Inkblot Test. Journal of Personality

Assessment. 50.(2), 222-228.

Naglieri, J. A. (1988). DAP Draw a person: A Quantitative scoring

system manual. San Antonio: The Psychological Corporation, Harcourt

Brace Jovanovich.

Naumberg, M. (1980). Art therapy: Its scope and function. In

E. F. Hammer (Ed.), The clinical application of projective

drawings (6th ed.) (pp. 511-517). Springfield, IL: Charles C.

Thomas.

Rabin, A. I. (1968) Projective methods: An historical Introduction.

In A. I. Rabin (Ed.), Projective techniques in personality

assessment (pp. 3-17). New York: Springer.

Reckase, M. D. (1984). Scaling Techniques. In G. Goldstein & M.

Hersen (Eds.). Handbook of Psychological Assessment. New York:

Pergamon Press.

Ryan, B. F., Joiner, B. L., & Ryan, T. A. (1985). Mlnltab handbook

(2nd ed.). Boston: PWS-Kent.

Schall, W. E., Silvestro, J. R., & Brown, M. (1985). Microcomputers

in the classroom: Don't exclude the developmentally disabled.

International Journal of Education and Computing. 1(3),

195-199.

Selmi, P. M., Klein, M. H., Greist, J. H., Sorrell, S. P., &

Erdman, H. P. (1990). Computer-administered cognitive-behavioral

therapy for depression. American Journal of Psychiatry. 141(1),

51-56.

Shaw, M. E., & Wright, J. M. (1967). Scales for the measurement of

attitudes. New York: McGraw-Hill.

Silva, D. (1986). Deluxe oaint II [Computer program]. San Mateo,

CA: Electronic Arts.

Steinberg, E. R. (1984). Teaching computers to teach. Hillsdale,

NJ: Lawrence Erlbaum Associates.

U.S. Congress, Office of Technology Assessment. (1988). Power

on! New tools for teaching and learning (OTA-SET-379).

Washington, D.C.: U.S. Government Printing Office.

Wildman, R. W., & Wildman, R. W. II (1975). An investigation into

the comparative validity of several diagnostic tests and test

batteries. Journal of Clinical Psychology. 11(3), 455-458.